From: KONRAD Frederic <fred.kon...@greensocs.com>
Hi everybody,
This is the seventh version of this RFC (see the changes below).
The first patches:
migration: make qemu_savevm_state public.
icount: introduce icount timer.
icount: check for icount clock deadline when cpu loop exits.
icount: make icount extra computed on icount clock as well.
are various preparation patches for reverse execution it creates the icount
clock.
The last patches:
trace-events: add reverse-execution events.
introduce reverse execution mechanism.
gdbstub: allow reverse execution in gdb stub.
cpu-exec: trigger a debug request when rexec stops.
rexec: synchronize icount on the next event.
rexec: allow to enable reverse execution.
are reverse execution introduction.
They can be clone at: git://git.greensocs.com/qemu_cexe.git:rexec_v7
This implementation of reverse execution works with instruction counting:
A new clock is implemented which is icount clock. It grows each time an
instruction is executed and is totally independant of host clock.
Snapshots are taken regularly (based on icount clock) with help of migration
code and written on the disk.
When user wants to use reverse-stepi:
* Last snapshot is reloaded.
* A stop callback is created to be triggered at the previous instruction.
This stop callback generates a debug exception so QEMU stops in debug mode.
Command line:
* rexec suboption is added to icount to enable reverse execution, it needs
icount=N and doesn't support auto mode.
About non determinism in QEMU:
* This implementation doesn't take IO in account so any IO will cause non
determinism and break reverse execution.
* The icount warp mechanism have been disabled when reverse execution is
enabled so the time grow differently inside the VM.
Testing:
* It has been tested on ARM without any IO such as network or asynchronous file
access to keep the deterministic behaviour of icount.
Known issues:
* On ARM stepi seems to do some additional steps which are added to icount
counter so reverse-stepi just after stepi is broken.
* The IO replay explained above.
Changes:
v6 -> v7:
* Rebase.
v5 -> v6:
* Boolize some variable (As suggested by Li Guang).
* s/snap_timer/snapshot_timer/g (As suggested by Li Guang).
* Choice a better name for the snapshot file. (As suggested by Li Guang).
* Boolize stepping_back, etc.. (As suggested by Li Guang).
* Cherry-pick icount: Add QemuOpts for icount. (As suggested by Paolo).
* Making reverse execution option using QEMUOpts (As sugested by Paolo).
* s/cexe/rexec/g (As nobody like cexe :)).
v4 -> v5:
* Fix a coding style issue (As suggested by Juan).
v3 -> v4:
* Fix icount_state_needed (As suggested by Amit).
* Rebase.
v2 -> v3:
* Use trace instead of debug printfs (As suggested by Lluis).
v1 -> v2:
* Use subsection for icount migration (As suggested by Paolo).
* Use with_bias parameters to get_icount instead of get_icount_wo_bias
function (As suggested by Paolo).
KONRAD Frederic (10):
migration: make qemu_savevm_state public.
icount: introduce icount timer.
icount: check for icount clock deadline when cpu loop exits.
icount: make icount extra computed on icount clock as well.
trace-events: add reverse-execution events.
introduce reverse execution mechanism.
gdbstub: allow reverse execution in gdb stub.
cpu-exec: trigger a debug request when rexec stops.
rexec: synchronize icount on the next event.
rexec: allow to enable reverse execution.
Makefile.target | 1 +
cpu-exec.c | 13 ++
cpus.c | 76 ++++++++--
gdbstub.c | 31 ++++-
include/qemu/timer.h | 17 ++-
include/reverse-execution.h | 43 ++++++
include/sysemu/sysemu.h | 1 +
main-loop.c | 10 ++
qemu-options.hx | 5 +-
qemu-timer.c | 8 +-
reverse-execution.c | 308 +++++++++++++++++++++++++++++++++++++++++
savevm.c | 2 +-
stubs/Makefile.objs | 1 +
stubs/cpu-get-icount.c | 10 +-
stubs/reverse-execution-stub.c | 32 +++++
trace-events | 6 +
vl.c | 16 ++-
17 files changed, 561 insertions(+), 19 deletions(-)
create mode 100644 include/reverse-execution.h
create mode 100644 reverse-execution.c
create mode 100644 stubs/reverse-execution-stub.c
--
1.9.0
